Background

The obspy framework is a set of Python libraries for observational seismology. It includes a number of routines to retrieve and process seismic data in a variety of formats.

Antelope is a commercial data acquisition and processing system for environmental data, with a heavy emphasis on seismic data. Many regional seismic networks use Antelope to acquire, process, and exchange seismic data. Starting with version 5.2, it has included Python bindings for many of it’s core routines for databases and real-time data.

The Obspy framework combined with Antelope represents a best-of-breed combination of interoperability with other seismic data formats and rapid application development. However, due to possible licensing restrictions, BRTT does not currently distribute Obspy.

BRTT ships their own distribution of Python with Antelope, so just installing the system packages for scipy and numpy included with RHEL is not an acceptable solution. You would not get access to the Antelope routines from Obspy installed in this fashion.

Obspy has a number of dependencies, some of which conflict with libraries shipped with Antelope. In particular, the Numpy is not configured with some required external dependencies such as BLAS and LAPACK that Scipy (another Obspy dependency) requires.

Thus, there are two ways to go about working around this issue:

Install a newer version of Numpy or overwrite the existing numpy library in the /opt/antelope/python2.7.6/site-packages directly
Install into a different site-packages directory outside of the BRTT-supplied directories.

The first approach, putting all of the new packages including this upgraded numpy module in the BRTT site-packages directory, introduces the possibility of compatibility issues with BRTT-supplied code (such as orbrtd) that may depend on their version of numpy (1.7.1). In practice there haven’t been any reported compatibility problems, but it’s best to not step on BRTT’s supported version of the software any more than you have to.

The second approach requires a bit more work on your end, but it ensures that BRTT-written programs will not have compatibility issues with any upgraded libraries.

My facility uses the second approach, but it’s tied to our own /opt/anf overlay to /opt/antelope. If you are sharing code with a number of users at a side, setting up an overlay tree for Antelope is a really good way to keep your code in revision control and provide it to all users on your system without modifying the BRTT /opt/antelope tree extensively. See the utility build_sourcetree in the Antelope contributed software repository on how to get off the ground with your own overlay.

For the purposes of this installation, I will assume that your site does not have a build_sourcetree-style overlay, and that you want obspy to be available to all users on your system. Thus, we will choose the directory /opt/antelope/local/lib/python2.7 as a place to install the obspy packages, and the binaries will go in /opt/antelope/local/bin.

Prerequisites

A working installation of Antelope 5.4 or later
… on a RHEL or CentOS 6 OS
… with the development tools installed (yum -y groupinstall “Development tools”)
Your shell environment must be set up with the Antelope environment - For Bourne-style shells, run source /opt/antelope/5.4/setup.sh

The instructions below assume you are using Bash or another Bourne-compatible shell. There’s no reason this wouldn’t work in CSH, but you’ll need to translate any environment variable commands to their Csh-equivalent yourself.

Installation

Step 0: Verify your Antelope environment is set up correctly

Run this command:

[[ -z "$ANTELOPE" ]] && echo "ANTELOPE ENVIRONMENT IS NOT SET CORRECTLY” || echo Antelope is OK

If that doesn’t come back with “Antelope is OK”, you need to source the Antelope environment

Now, run this command:

which easy_install | grep antelope || echo "You're not using the right easy_install"

If that complains that you’re not using the right easy_install, your path may be messed up. Make sure that the Antelope Python path comes before any other PATH statements.

Step 1: Install system library dependencies

yum install -y lapack-devel blas-devel

Step 2: Create your local Python package tree and add it to your PYTHONPATH temporarily

pymoddir=/opt/antelope/local/lib/python$(getid python_mainversion)
bindir=/opt/antelope/local/bin
export PYTHONPATH=$pymoddir
mkdir -p $pymoddir
mkdir -p $bindir
export EASY_INSTALL_ARGS="-d $pymoddir -s $bindir -N"
unset pymodir
unset bindir

There’s a couple of things going on in that blob of shell commands that are worth noting. First is the usage of the Antelope command getid. This program allows you to get a bunch of configuration information from Antelope, including the Python version in use. Run the command getid -a for a list of all of the information that getid can report.

Note that we could have also gotten the python major and minor version (python_mainversion above) by running a python one-liner: python -c 'import sys; sys.version[:3]'

Secondly, the EASY_INSTALL_ARGS variable is being set just to save us some typing in later commands.

Finally, unless we set PYTHON_PATH temporarily, easy_install will complain that the directory we are trying to install modules into is not part of the Python search path and will refuse to run.

Step 3: Install numpy

The latest version of numpy as of this writing (1.8.0) has a bug with it’s f2py code that prevents Scipy from working properly. Use 1.7.2 instead.

If all goes right, the numpy installer will find BLAS and LAPACK and link against them. You can safely ignore any warnings about Atlas, unless you feel that you will need the Atlas routines.

easy_install ${EASY_INSTALL_ARGS} numpy==1.7.2

Step 4: Install scipy

Scipy’s installer bugs out if you don’t explicitly set the CC and CXX variables.

CC=/usr/bin/gcc CXX=/usr/bin/g++ easy_install ${EASY_INSTALL_ARGS} scipy

Step 5: Install a couple of other dependencies

You’ll also need lxml, suds, and sqlalchemy.

easy_install ${EASY_INSTALL_ARGS} lxml
easy_install ${EASY_INSTALL_ARGS} suds
easy_install ${EASY_INSTALL_ARGS} sqlalchemy

Step 6: Install obspy itself

easy_install ${EASY_INSTALL_ARGS} obspy

Usage

In order to actually use the obspy libraries in code, you’ll need to remember to include the /opt/antelope/local tree in your Python search path. There are a couple of ways to make this work. The best way (which ensures that nothing weird will happen with core BRTT programs) is to explicitly modify the python search path in your code itself.

If you set your PYTHONPATH to /opt/antelope/local/lib/python2.7, things will probably work but you run the risk of odd behavior with core Antelope programs.

Thus, it’s recommended that each program is prefixed with a line like

import site; import sys; site.addsitedir('/opt/antelope/local/lib/python' + sys.version[:3])

It’s important to use site.addsitedir instead of sys.path.append because the latter doesn’t evaluate easy_install.pth, and thus Python won’t see any of the new modules you installed.

In `iPython` or as a standalone script

A full pasteable blurb that should get iPython ready to use Obspy and Antelope looks like this:

``` python Paste this into a standalone script or iPython import os import sys import site

import signal

signal.signal(signal.SIGINT, signal.SIG_DFL)

sys.path.append(os.environ[‘ANTELOPE’] + “/data/python”) site.addsitedir(‘/opt/antelope/local/lib/python’ + sys.version[:3]) ```

As an `.xpy` file

If you use the ANTELOPEMAKE structure to build Antelope programs, there is afile type called .xpy that will configure your script with a preamble similar to the iPython blurb above, except it does not include the last line.

Every xpy file that makes use of the obspy code will need the following line pre-pended to the file before any obspy module imports are made (Not that the standard xpy header already imports os and sys for you):

site.addsitedir('/opt/antelope/local/lib/python' + sys.version[:3])

The following example.xpy will get “compiled” into a script named example:

``` python example.xpy site.addsitedir(‘/opt/antelope/local/lib/python’ + sys.version[:3])

import numpy as np import matplotlib.pyplot as plt from obspy.core import UTCDateTime from obspy.arclink import Client from obspy.signal import cornFreq2Paz, seisSim

client = Client(user=’test@obspy.de’)

t = UTCDateTime(‘2009-08-24 00:20:03’) st = client.getWaveform(‘BW’, ‘RJOB’, ‘’, ‘EHZ’, t, t + 30) pas = client.getPAZ(‘BW’, ‘RJOB’, ‘’, ‘EHZ’, t)

1Hz instrument

one_hertz = cornFreq2Paz(1.0)

Correct for frequency response of the instrument

res = res / paz[‘sensitivity`]

Plot the seismograms

sec = np.arange(len(res)) /st[0].stats.sampling_rate

plt.subplot(211) plt.plot(sec, st[0].data, ‘k’) plt.title(“%s %s” % (st[0].stats.station, t)) plt.ylabel(‘STS-2’)

plt.subplot(212) plt.plot(sec, res, ‘k’) plt.xlabel(‘Time [s]’) plt.ylabel(‘1Hz CornerFrequency’)

plt.show() ```